FYP Project Ideas for University Students

Real time Suspicious Behavior Detection using Big Data Tools and Techniques:

In this project Graphical application will be implemented for real time detection of suspicious behaviors. The surveillance system for suspicious behavior detection has very high importance. It is used for the concept of video analytics. The purpose of this is to ensure a high level of security in a public places. Understanding people behaviors in real time allows the surveillance systems to analyze unusual events through the video frames. The development of Surveillance System includes ”Managing the video streaming data” and “Analyzing Video Frames (Image Processing)”. Following steps must be followed for analyzing video frames: • Converting video in frames at real time. • Analyzing each frame by detecting the objects. • SIFT (scale-invariant feature transform) Algorithm will be used for object recognition, mapping and navigation, image stitching, gesture recognition and match moving. • Video Frame analysis also include the vector orientation and the displacement of the object. The variation in vector orientation and displacement of object from frame to frame allows us to identify whether the object mobility is suspicious or not. • If the moving object is caught suspicious then the surveillance system should produce the alert on the user interface.

Plants phenotypic traits using ML :

Motivation: A deeper understanding of the biological processes mediated by plant genomes is needed to develop crops with improved stress resilience and yield potential. Connecting genotype to phenotype for quantitative plant traits on a genome level necessitates high-density genetic markers and large population sizes to gain sufficient power and resolution. Better phenotype analysis the more likely that new genes and complex interactions will be revealed. Scope: The purpose of the project is develop a platform for deep convolutional neural networks for the plant phenotyping. Complex leaf phenotyping traits will be used as a benchmark. Convolutional neural networks (CNNs) will be constructed and trained to detect benchmark traits. CSCL phenotypic dataset will be used to train and test platform in first phase, furthermore, it will be used to detect benchmarked phenotypic traits of plant using data collected from a local (CSRL) agriculture site.